Author Archives: ryanchisholm

New multi-authored article on the history of the ForestGEO network published in Biological Conservation

In recent years, our lab has played a lead role in several cross-site analyses of data from the global ForestGEO (formerly CTFS) network (see here, here, here, and here). In a new article just published in Biological Conservation, Stuart Davies, the director of the ForestGEO network, relates the history of the network and summarises some of the main scientific results emerging from it. Ryan is among the 100+ coauthors of the article.

Davies, S. J., I. Abiem, K. Abu Salim, S. Aguilar, D. Allen, A. Alonso, K. Anderson-Teixeira, A. Andrade, G. Arellano, P. S. Ashton, P. J. Baker, M. E. Baker, J. L. Baltzer, Y. Basset, P. Bissiengou, S. Bohlman, N. A. Bourg, W. Y. Brockelman, S. Bunyavejchewin, D. F. R. P. Burslem, M. Cao, D. Cárdenas, L.-W. Chang, C.-H. Chang-Yang, K.-J. Chao, W.-C. Chao, H. Chapman, Y.-Y. Chen, R. A. Chisholm, C. Chu, G. Chuyong, K. Clay, L. S. Comita, R. Condit, S. Cordell, H. S. Dattaraja, A. A. de Oliveira, J. den Ouden, M. Detto, C. Dick, X. Du, Á. Duque, S. Ediriweera, E. C. Ellis, N. L. E. Obiang, S. Esufali, C. E. N. Ewango, E. S. Fernando, J. Filip, G. A. Fischer, R. Foster, T. Giambelluca, C. Giardina, G. S. Gilbert, E. Gonzalez-Akre, I. A. U. N. Gunatilleke, C. V. S. Gunatilleke, Z. Hao, B. C. H. Hau, F. He, H. Ni, R. W. Howe, S. P. Hubbell, A. Huth, F. Inman-Narahari, A. Itoh, D. Janík, P. A. Jansen, M. Jiang, D. J. Johnson, F. A. Jones, M. Kanzaki, D. Kenfack, S. Kiratiprayoon, K. Král, L. Krizel, S. Lao, A. J. Larson, Y. Li, X. Li, C. M. Litton, Y. Liu, S. Liu, S. K. Y. Lum, M. S. Luskin, J. A. Lutz, H. T. Luu, K. Ma, J.-R. Makana, Y. Malhi, A. Martin, C. McCarthy, S. M. McMahon, W. J. McShea, H. Memiaghe, X. Mi, D. Mitre, M. Mohamad, L. Monks, H. C. Muller-Landau, P. M. Musili, J. A. Myers, A. Nathalang, K. M. Ngo, N. Norden, V. Novotny, M. J. O’Brien, D. Orwig, R. Ostertag, K. Papathanassiou, G. G. Parker, R. Pérez, I. Perfecto, R. P. Phillips, N. Pongpattananurak, H. Pretzsch, H. Ren, G. Reynolds, L. J. Rodriguez, S. E. Russo, L. Sack, W. Sang, J. Shue, A. Singh, G.-Z. M. Song, R. Sukumar, I. F. Sun, H. S. Suresh, N. G. Swenson, S. Tan, S. C. Thomas, D. Thomas, J. Thompson, B. L. Turner, A. Uowolo, M. Uriarte, R. Valencia, J. Vandermeer, A. Vicentini, M. Visser, T. Vrska, X. Wang, X. Wang, G. D. Weiblen, T. J. S. Whitfeld, A. Wolf, S. J. Wright, H. Xu, T. L. Yao, S. L. Yap, W. Ye, M. Yu, M. Zhang, D. Zhu, L. Zhu, J. K. Zimmerman, and D. Zuleta. 2021. ForestGEO: Understanding forest diversity and dynamics through a global observatory network. Biological Conservation 253:108907.

New review paper on conspecific negative density dependence and tree diversity published in Trends in Ecology and Evolution

We have just published a review paper, led by Lisa Hülsmann of the University of Regensburg, about conspecific negative density dependence and its ability to explain tree diversity. The predominant pattern in global tree diversity is increased species richness towards the tropics. One proposed explanation for this is the greater climatic stability of tropical forests, which allows greater prevalence of pests (e.g., herbivorous insects and fungi), which in turn keep the abundances of their host tree species in check, thus maintaining overall tree diversity. For this mechanism to work, a pest must have greater per-tree impacts when the host tree is at high population density. This is an example of a more general phenomenon called conspecific negative density dependence.

The idea that pests maintain the tree diversity of tropical forests was proposed 50 years ago by Daniel Janzen and Joseph Connell and eventually became known as the Janzen–Connell hypothesis. In the years since, many empirical studies have reported that tree species do suffer more when surrounded by individuals of their own species, consistent with the hypothesis. These observations have provoked optimism among forest ecologists that the Janzen–Connell hypothesis is close to proven.

In our review, we present a more cautious appraisal. Our summary of the current state of knowledge reveals two important unresolved questions. Firstly, it is not clear whether the effect of neighbouring conspecific trees is strong enough to have a substantial influence on the overall tree diversity in a forest. Secondly, it is not yet possible to say whether the regulatory effect is indeed stronger or more frequent in the tropics.

We conclude that the explanation of Janzen and Connell remains a hypothesis yet to be proven. More precisely, although the existence of the mechanism is relatively well established, its importance in comparison to many other alternative explanations for tropical tree diversity remains unclear. To weigh these hypotheses against each other and to test the Janzen–Connell hypothesis in its entirety, new data and collaborations between experimental and theoretical ecologists will be necessary.

We hatched the idea for this review during a visit to Florian Hartig‘s lab in Regensburg in 2018.

Hülsmann, L., R. A. Chisholm, F. Hartig. 2020. Is variation in conspecific negative density dependence driving tree diversity patterns at large scales? Trends in Ecology and Evolution (in press)

Deon’s paper on comparing undescribed extinction models published in Conservation Biology

In recent years, the lab has been working on how to account for undescribed species when estimating extinction rates (see previous posts here, here, and here). The issue at stake is that some species may go extinct before they become known to science, and that failure to account for these statistically can lead to underestimates of extinction rates. In a new paper led by Deon, we compare our lab’s SEUX model with an earlier model by Tedesco et al. (2014). For the new paper we collaborated with Tedesco and tested the two models against his original global data sets as well as simulated data.

Reassuringly, the two models produced fairly similar estimates of the proportion of extinct species, accounting for undescribed species, when applied to the global data sets (see figure below). For example, the SEUX and Tedesco models estimated, respectively, that 12% and 11% of Australian mammals have gone extinct over the last 200 years, compared to the naïve estimate of 7% (which ignores undescribed extinctions). However, a few caveats emerged. Firstly, the SEUX model assumes that there are no extant undescribed species in the present day, and as a result of this assumption may underestimate the absolute number of extinctions (as opposed to the proportion). Secondly, both models assume that the probability of a species going extinct is independent of its probability of being described. Applications to our simulated data showed that violation of this latter assumption can lead to large biases in the extinction estimates. We also found evidence that the assumption may be violated in a few of our real-world data sets. More work is needed to investigate possible correlations between extinction and detection rates. Despite these caveats, our two models reinforce the notion that undescribed extinctions can be large and need to be accounted for in holistic assessments of human impacts on the environment.

Lum, D. W. H., P. A. Tedesco, B. Hugueny, X. Giam, and R. A. Chisholm. 2020. Quantifying the relative performance of two undetected-extinction models. Conservation Biology (in press)

The two undescribed extinction models (Tedesco and SEUX) produce similar estimates of the proportion extinct species for most data sets.

New paper on Janzen–Connell effects published in The American Naturalist

The high diversity of tropical plant communities is an enduring mystery, and a range of hypotheses have been proposed to explain it. In a new paper, Tak and Ryan evaluate the plausibility of the prominent Janzen–Connell hypothesis, which proposes that tropical plant diversity is maintained by selection against common species via impacts of natural enemies. Using a modelling approach informed by data from tropical forests in Panama, they find generally weak support for the hypothesis, with high diversity being maintained only in the idealised case where all species are equivalent. The paper has just come out online at The American Naturalist, and will be published in the forthcoming November issue. The lay summary and Abstract from the journal website are reproduced below.

R. A. Chisholm & T. Fung. 2020. Janzen–Connell effects are a weak impediment to competitive exclusion. The American Naturalist (in press)

UPDATE (December 2020): Our paper has been recommended on Faculty Opinions (formerly F1000).

Rainforest canopy crane at Fort Sherman, Panama.

A tropical forest, such as this one at Fort Sherman, Panama, can contain over 100 tree species within a single hectare. (Marcos A. Guerra/Smithsonian Tropical Research Institute)

Are pests and pathogens responsible for high tree diversity in tropical forests?

Step into a tropical forest and before long something will attack you. A mosquito bites you, a tick latches on, a chigger bores under your skin. You sit on the ground and get a nasty case of worms. Eventually, fungi eat away at your clothes. Such pests and pathogens can make life difficult—not just for humans but for plants and other tropical forest organisms as well. But could these “natural enemies” in fact be essential to the ecosystem, by preventing any one species from getting the upper hand? In a new study, two ecologists from the National University of Singapore tested this idea for tropical forest trees by building mathematical models in which each tree species is attacked by its own specialised enemy. In real tropical forests, there can be dozens or hundreds of tree species in a single hectare. Did their model produce similar diversity? The answer in general was no. Only in a very special case, the ecological equivalent of a pin landing on its head, did they e get such high tree diversity. This special case was when all species were equivalent—they all had the same death rate, same birth rate, and so on. Because pins generally do not land on their heads in nature, this scenario is implausible. The idea that pests and pathogens, through their voracious activities, can maintain forest diversity is vivid and appealing, but it does not stand up to scrutiny. The quest for a rigorous explanation of tropical forest diversity continues.


A goal of ecology is to identify the stabilizing mechanisms that maintain species diversity in the face of competitive exclusion and drift. For tropical forest tree communities, it has been hypothesized that high diversity is maintained via Janzen–Connell effects, whereby host-specific natural enemies prevent any one species from becoming too abundant. Here, we explore the plausibility of this hypothesis with theoretical models. We confirm a previous result that when added to a model with drift but no competitive exclusion, i.e., a neutral model where intrinsic fitnesses are perfectly equalized across species, Janzen–Connell effects maintain very high species richness that scales strongly with community size. However, when competitive exclusion is introduced, i.e., when intrinsic fitnesses vary across species, the number of species maintained by Janzen–Connell effects is substantially reduced, and scales much less strongly with community size. Because fitness variation is pervasive in nature, we conclude that the potential of Janzen–Connell effects to maintain diversity is probably weak, and that the mechanism does not yet provide a sufficient explanation for the observed high diversity of tropical forest tree communities. We also show that, surprisingly, dispersal limitation can further reduce the ability of Janzen–Connell effects to maintain diversity.

Sam’s paper presenting his neutral coalescence simulation packages published in Methods in Ecology and Evolution

Computational methods for efficiently simulating ecological communities are increasingly in demand. In 2008, James Rosindell popularised coalescence methods for simulating neutral ecological communities. These methods can be orders of magnitude faster than the alternatives, and can make spatially explicit neutral simulations on an effectively infinite landscape feasible. However, the technical challenge of implementing coalescence methods has left them out of reach of many ecologists. Now, Sam Thompson, who recently graduated from the lab with a PhD, being co-supervised by Rosindell at Imperial College London, has published user-friendly high-performance software packages for performing coalescence simulations in R and Python. The packages are described in a new paper in Methods in Ecology and Evolution.

The essence of coalescence methods is that they run backwards in time, which increases efficiency because only past lineages that lead directly to individuals within the focal community in the present day need to be tracked. In contrast, with standard forward-time-simulations, a much larger number of lineages needs to be tracked because it is not known in advance which ones will persist and be part of the focal community in the future. The catch is that coalescence can only be applied to particular types of ecological models, in particular to neutral models where an individual’s chances of birth and death are independent of its species identity. Coalescence techniques were originally invented in population genetics, and then imported to ecology by Rosindell.

Sam’s new software packages include the basic features from Rosindell’s 2008 paper, along with many enhancements and novel features, including varying individual density across a landscape, changing landscape configuration over time, and a variety of dispersal and speciation modes. The software was also used in Sam’s recent Ecology Letters paper on extinction debt. The core package code is written in C++, but the R and Python packages give a user-friendly interface, that should enable easy uptake by any ecologists interested in running fast spatially explicit neutral simulations.

Samuel E. D. Thompson, Ryan A. Chisholm & James Rosindell. pycoalescence and rcoalescence: Packages for simulating spatially explicit neutral models of biodiversity. Methods in Ecology and Evolution (in press)

Screen Shot 2020-08-17 at 4.58.44 pm

Output from the new software packages, showing (c) change in species richness over time in a hypothetical scenario after (d) part of a landscape surrounding a national park (green area) has been cleared.

Nadiah’s new paper on Singapore plant extinctions out in Conservation Biology

Have you ever wondered how many species were lost before we had the chance to discover them? In a paper now out in Conservation Biology, led by Nadiah, we estimated just that, for plant species in Singapore, following on from our lab’s related work on Singapore birds and butterflies.

All over the world, many species remain undiscovered while both known and unknown species continue to go extinct. This is particularly true in the tropics, where biodiversity is high and development continues apace. Singapore provides an invaluable case study of tropical biodiversity loss. Since British colonisation in 1819, most of Singapore’s forest cover has been replaced with urban landscape. However, Singapore also has one of the best-documented floras in the world, in terms of taxonomic and temporal coverage, with historical collections beginning only a few years after colonisation.


The orchid species Grammatophyllum speciosum has not been recorded in Singapore since 1918.
Photo credit: Cerlin Ng

We collated a high-quality database of over 30,000 Singapore plant collections representing over 2,000 species, and we applied our lab’s previously developed “SEUX” model to estimate extinction rates and total numbers over time. The SEUX model is based on a straightforward idea: if we assume that the per-year per-species extinction rates have been the same for discovered and undiscovered extinctions, then we have a basis for working backwards in time to estimate the number of undiscovered species and the proportion that went extinct. In the new paper we also developed a more accurate method for obtaining confidence intervals on the estimates.

We estimated that 30–38% of Singapore plant species have gone extinct since 1819. The central estimate using classical methods was 32% and that using Bayesian methods was 35%. Crucially, these numbers are much higher than the 22% extinction rate that one obtains from the naïve method of simply dividing the number of known extinctions by the total number of discovered species, demonstrating the importance of calculating extinction rates in a way that accounts for unknown species.

Also check out Nadiah’s blog post on the paper, her previous blog post giving a SEUX tutorial, and her SEUX for R package on GitHub. The full reference for the paper is below.

Kristensen, N. P., Seah, W. W., Chong, K. Y., Yeoh, Y. S., Fung, T., Berman, L. M., Tan, H. Z., Chisholm, R. A. (2020) Extinction rate of discovered and undiscovered plants in Singapore, Conservation Biology (in press)


From the Singapore plants database, we inferred the number of discovered species that were extant and extinct over time. We then used the SEUX model on these data to estimate the number of undiscovered species that were extant and extinct over time.


Aloysius’s paper on estimating tropical plant litter decomposition published in Pedobiologia

Aloysius Teo completed his PhD in the lab in 2017, and a chapter from his thesis has just been published in Pedobiologia. The “tea bag method” was developed several years ago by our coauthor Joost Keuskamp for estimating decomposition rates of plant litter. The method relies on commercially available tea bags, thus facilitating standardisation of methods across studies. Aloysius explored the applicability of the method to tropical forests, where a particular problem is that abundant termites readily damage tea bags.

Aloysius found that tea bag attack rates by termites were large enough in tropical forests in Singapore to invalidate experimental results relying on the tea bag method. He trialed methods for excluding termites and, based on his results, recommended an extended tea bag method for future use in the tropics that relies on a combination of unmodified tea bags and termite exclusion treatments.

Teo, A., N. P. Kristensen, J. A. Keuskamp, T. A. Evans, M. Foo, R. A. Chisholm. 2020. Validation and extension of the tea bag index to collect decomposition data from termite-rich ecosystems. Pedobiologia (in press)

tea bags

Top: Physical termite exclusion barriers used for tea bags in the study, along with unmodified tea bags. Bottom: A tea bag undetected by termites (left) alongside two bags that were detected (centre and right).


Tak coauthors new genetic modelling paper with Frank Rheindt’s lab

A new paper led by Tang Qian and  Frank Rheindt from the Avian Evolution Lab at NUS, with Tak Fung as a co-author, has just been published in Molecular Ecology Resources.

The paper describes how they developed a new R package called ResDisMapper, which helps in the management of biological invasions and habitat degradation by allowing users to generate a map showing resistance to dispersal over a landscape, as defined using genetic data. The R package is novel because there are few programs available that map resistance to dispersal over the relatively short spatiotemporal scales required for the management of biological invasions and habitat degradation.

They tested ResDisMapper against two other programs (DResD and EEMS) using a suite of simulated datasets and found that overall, it performed substantially better. They further demonstrated the utility of ResDisMapper by applying it to genetic data collected for rock pigeons (Columbia livia) in Singapore and Golden-crowned sifakas (Propithecus tattersalli) in northern Madagascar, to identify regions with high and low resistance to dispersal.

Tang, Q., T. Fung, and F. E. Rheindt. 2020. ResDisMapper: An R package for fine-scale mapping of resistance to dispersal. Molecular Ecology Resources


Resistance map produced by ResDismapper for rock pigeons in Singapore, with annotations describing the meaning of the different colours and contours. A significant barrier/corridor refers to areas with resistance values that are higher/lower than those from a null distribution with high probability, and lie within the red/green contours. Areas that lie inside the blue contours have resistance values with high probability of being positive or negative (high “certainty”). The yellow circles indicate sampling points.